Re: Getting a LineOfDevelopment object from branch name

Greg Ward wrote:
> in HgOutputOption and related code, I occasionally need to dig up the
> Trunk instance used to represent CVS trunk in this conversion, or the
> Branch object used to represent some CVS branch.  I'm handling it by
> keeping a dict from branch name (or "trunk") to LineOfDevelopment
> object, which I populate in process_initial_commit() and
> process_branch_commit().  This seems fine, but I was wondering if
> there is an easier way -- did I miss some global mapping of
> LineOfDevelopment objects?

Since, when converting to Subversion, cvs2svn can handle multi-project
conversions, there is a separate Trunk instance for each project.  Its
ID is stored in Project.trunk_id during CollectRevsPass, and the Trunk
object itself can be retrieved using Project.get_trunk().

All LinesOfDevelopment, including the Trunk objects and ExcludedSymbols,
are available from Ctx()._symbol_db during most passes (including
OutputPass).  This is a mapping from ID -> LOD object.  But everything
is in memory, so it is cheap to iterate over all symbols:

symbol_map = {}
for symbol in Ctx()._symbol_db:
  if isinstance(symbol, (Branch, Tag)):
    symbol_map[symbol.name] = symbol
  elif isinstance(symbol, Trunk):
    symbol_map[SPECIAL_TRUNK_KEY] = symbol

Please note that Symbol.name contains the symbol name *after* any symbol
transforms, which is not necessarily the same as the original CVS symbol
name.  (Transformed symbol names and CVS symbol names are not even
necessarily 1:1.)  The pre-transformed name is not recorded anywhere.

Hope that helps,
Michael

------------------------------------------------------
http://cvs2svn.tigris.org/ds/viewMessage.do?dsForumId=1667&dsMessageId=2385004

To unsubscribe from this discussion, e-mail: [[email protected]].